IVRにおけるAIの活用

Applications of artificial intelligence in interventional radiology

近年の人工知能（artificial intelligence：AI）技術の急速な発展には目を見張るものがあり，医療分野においても例に漏れずAIとの融合が大きな注目を集めている。interventional radiology（IVR）においては，深層学習を中心とするAIの飛躍的発展により，読影支援，病変検出，術中画像処理，予後予測といったIVRに係る各フェーズでAIが徐々に実用段階に入り，IVRの刷新が始まっている。本稿では，IVRにおけるAIの活用をテーマに，過去のコンピュータ支援診断（computer-aided diagnosis：CAD）の歴史から最新の深層学習の応用技術まで，幅広い観点から詳述する。

Recent advances in artificial intelligence（AI）, especially deep learning, are reshaping every phase of interventional radiology（IR）. This article traced the evolution from early computer-aided diagnosis aimed at full automation to today’s AI systems that augment physicians in lesion detection, image processing, and outcome prediction. We summarized core machine-learning techniques, including convolutional neural networks, and discussed emerging approaches such as radiomics, radiogenomics, and AI-guided robotics. The current challenges in the clinical application of AI are data heterogeneity, explainability, regulatory hurdles, and ethical issues. Understanding and integrating AI responsibly will be essential for IR physicians to deliver safer, more personalized, minimally invasive care.